Key Takeaways

  • G2's 2026 Buyer Behavior Report found that 82% of B2B software buyers report a specific AI-driven shift in how they research and purchase software, and roughly half now start with an AI chatbot before touching a vendor site.
  • The buying journey is fragmenting into machine-mediated research (chatbots, agents, LLM answers) and human-mediated validation (reviews, peer signals, sales conversations), and marketers need to show up in both.
  • Review platforms and structured, extractable content are now feeding the AI answers your buyers see, which means G2, Gartner Peer Insights, and your own site's schema are compounding marketing assets, not check-the-box work.
  • The old MQL funnel is being replaced by a smaller number of higher-intent conversations, because buyers arrive already shortlisted by an LLM, and your sales motion has to assume the buyer knows more than you think.
  • The wrong response is chasing every AI channel at once. The right response is picking two or three where your actual buyers are researching, and instrumenting them properly.

A CMO friend forwarded me the G2 post last week with one line above it: "is this real?" The number that stopped her was 82%. That's the share of B2B software buyers in G2's June 2026 survey who say AI has already changed a specific step in how they buy software. Not "will change." Changed.

My read is that this is the first piece of first-party buyer data I've seen that lines up with what we're watching happen inside client accounts. Buyers are showing up to first calls with a shortlist they built with ChatGPT, they're quoting review snippets we didn't know existed, and they're asking questions a chatbot fed them rather than questions a salesperson coached them into. If you're a CMO and your pipeline reporting still assumes the buyer starts on your homepage, the map you're using no longer matches the terrain.

So this post is a walk through what the G2 numbers actually say, what they line up with across the other 2026 buyer studies, and the one piece of the consensus reading I think most CMOs are getting wrong.

What G2's 2026 report actually found

G2 surveyed more than 1,000 B2B software buyers in June 2026, and 82% of them reported a specific AI-driven change in their buying process. That's the headline. The more useful number sits in a companion piece from Demand Gen Report a month earlier: 51% of B2B buyers now start their research with an AI chatbot rather than a search engine or a vendor site, which is a threshold worth sitting with, because it isn't a signal of interest or experimentation, it's the actual starting line of the journey.

If half of your buyers begin inside a chat window you don't control, then the first impression of your category, your product, and your competitors is being written by a model working from whatever it can find, index, and summarize. Your careful homepage hero is showing up, at best, on visit three.

G2's earlier July 2025 report already flagged the direction: two out of three B2B software buyers actively consider AI capabilities when selecting software, and 88% of power users say they'd pay more for it. The 2026 update tells us the buyers who evaluate AI in products are now using AI to do the evaluating. The tool and the method converged.

The two-layer buying journey, in plain terms

Here's the way I'd describe the shift to a board. The old buying journey was a funnel we could see: awareness content, a form fill, an MQL, a demo, a proposal. The new journey has two layers stacked on top of each other. Underneath, there's a machine layer, chatbots, LLM answers, AI-augmented review sites, and increasingly, buying agents that summarize and shortlist on the buyer's behalf. On top, there's a human layer, the actual conversations, the peer references, the Slack DM to a friend who uses the tool. What changed is that the machine layer runs first and does more of the filtering than we realize.

Semrush's survey of B2B buyers found the same pattern from a different angle: AI both helps buyers discover brands they hadn't heard of and shapes the final vendor decision, not just the early research. Adobe's research on B2B customer experience in an AI-driven world reaches a similar place, noting that meaningful buying touchpoints are increasingly happening off the vendor's own properties, on review sites, in AI answers, in communities, in podcasts. The Geisheker analysis is more blunt: vendor selection is now happening before the first sales contact in most deals.

Which means the thing marketers have quietly known for two years is now measurable. By the time a buyer talks to your SDR, the shortlist is set. Your job is to be on it.

Where this actually hits your marketing plan

If the shortlist is being set before the first call, then the marketing questions worth asking all point to the same place: what shapes the shortlist, and are you visible there. I'd work through three moves in order, because they build on each other.

The first move is treating your review presence as infrastructure. G2, Gartner Peer Insights, TrustRadius, Capterra are training data now, not lead sources. When a buyer asks Claude or ChatGPT "who are the top vendors for X," the answer is being assembled partly from what those platforms say about you and your competitors. If you have 47 reviews and your competitor has 312, the model is going to lean toward the one with more signal. If your reviews are three years old, they'll get discounted for freshness. This isn't a hunch, it's how retrieval works. The work here is unsexy: run a real review generation program, keep it going every quarter, and make sure the language in those reviews reflects the specific problems your ICP searches on, use cases, integrations, workflows, the kind of specific claim a model can actually pull.

Which is what makes the second move follow so naturally: once the reviews are feeding the models, your own site has to be legible to those same models. The way I'd test whether your site is AI-ready is to ask Claude or ChatGPT a category question and see whether your product comes up, and if it does, whether the description matches how you'd actually describe yourself. If the answer is "no" or "sort of," it's usually because your pages are heavy on brand voice and light on structured, extractable facts. The move is to add proper schema, write FAQ blocks that answer the questions buyers actually type into chatbots, and keep a plain-language product page that lists what the thing does, who it's for, what it integrates with, and what it costs to run. The brand storytelling still matters for the human layer, but the machine layer wants the specs.

The third move is the one nobody wants to make, because it lives in sales, not marketing. Half your buyers are showing up having done research inside an AI tool that told them what your product does, what it costs (approximately), and how it compares to two competitors. Your discovery script that assumes ignorance is now insulting. The Demand Spring piece from earlier this year makes the point well: AI agents compress the manual research and make the human buyer faster and more prepared, which means the reps who are winning right now are the ones who ask "what have you already learned?" first, and adjust. The ones who are losing are still doing the 20-minute company overview slide.

Where I think most CMOs are reading the 82% wrong

The consensus reading of G2's number goes something like this: AI is now in the buying process, so we need an AI channel strategy, so let's chase every surface, chatbots, agents, LLM optimization, AI-native ad networks. That reading treats the 82% as a mandate to expand.

I'd bet against it. The more useful reading is that the 82% is a mandate to concentrate. If buyers are being shortlisted by models trained on a narrow set of high-signal sources, the leverage is in dominating those sources, not in showing up thinly across ten new ones. A category where two review platforms and your own structured site pages feed 80% of the model answers doesn't reward a scattered presence, it rewards depth in the three places that actually get retrieved. Most teams reading G2's report this quarter are going to add surfaces. The teams that win are going to subtract.

When this is the wrong thing to focus on

I want to be honest about where this doesn't apply. If you're selling a $2M enterprise platform into a Fortune 100 buying committee, the AI chatbot layer matters less than the analyst relationships, the executive references, and the RFP process. The G2 data skews toward the SMB and mid-market SaaS buyer, and that's where the shift is most violent. If your average deal is a two-year procurement cycle with a formal sourcing team, don't rip up your plan.

And if you're a small marketing team already stretched thin, don't try to chase every AI surface at once. Pick two, usually review platforms and LLM-visible content on your own site, and do those well before you worry about buying agents or AI-native ad channels. There's a version of this post that ends with "transform everything," and that version is wrong. Most teams should do fewer things, more deliberately. If you want a structured way to sequence that, we put ours in the Mighty & True blueprint.

What I'd do this quarter if I were you

Start with the LLM test on your top ten category queries, write down where you show up, where you don't, and what the models get wrong about you, and let that become the content brief for the quarter. From there, audit review velocity on the two platforms that matter most in your category and set a quarterly quota to keep them fresh. Then sit in on three sales calls and listen for what buyers already know before the rep starts talking, because that gap between what they know and what your rep assumes is the real cost of the shift G2 is measuring.

The 82% number will get quoted a lot over the next few months. The question I'd put to my team on Monday is simpler: which two surfaces are we going to own, and what are we cutting to do it.

Frequently Asked Questions

What percentage of B2B buyers use AI in their buying process?

According to G2's 2026 Buyer Behavior Report, 82% of B2B software buyers report a specific AI-driven change in how they buy software, and Demand Gen Report's coverage of the same research found 51% now start their research with an AI chatbot rather than a search engine or vendor site.

Does AI actually influence the final vendor choice, or just early research?

Both. Semrush's survey found AI shapes both brand discovery and the final vendor decision, and G2's data shows two out of three buyers now consider a product's own AI capabilities when selecting software. The idea that AI only touches the top of the funnel is out of date.

Should we still invest in traditional SEO if buyers are using AI chatbots?

Yes, because most LLM answers are still assembled partly from indexed web content, review sites, and structured data. What changes is that ranking is no longer the goal by itself. The goal is being retrievable, quotable, and factually clear enough that a model can pull a specific claim about you into an answer.

How do we get our product mentioned in ChatGPT or Claude answers?

Start by running the queries your buyers actually ask and see what comes back. Then work on the inputs the models are reading: fresh, specific reviews on the platforms in your category, structured content on your own site, schema markup, and factual FAQ pages. There's no ad slot to buy, so the work is upstream in what the models can find and trust.

Is the traditional MQL funnel dead?

The funnel is shrinking rather than disappearing. G2's data and Adobe's research both suggest buyers arrive further along the journey, with fewer form fills and more self-directed research. Expect fewer, higher-intent leads and a sales motion that has to assume the buyer already did the homework.

Sources